The foundational thesis of the last decadeโ€”that software is eating the worldโ€”has been inverted. In a recent essay, Theo Priestley argues that AI is now eating software, effectively stripping away the competitive moats that once protected SaaS valuations. The core premise is brutal for builders: if generative models can decompile, clone, and remix applications before they even hit full release, the economic value of proprietary code collapses. This isn't just theoretical; the author notes that 'there's not a piece of software that can't be decompiled, copied, made free, remixed, or cloned' in the current landscape.

The Commoditization of Code

For infrastructure engineers and product teams, the implication is clear: the barrier to entry has dropped to near zero. Priestley describes a 'speeding and out of control train' of AI investment, where VCs are laying down tracks to keep the momentum going despite a lack of tangible returns. The sunk cost fallacy is driving this loop. As money pours into AI tools and infrastructure, those very tools become vulnerable to being copied by competitors in 'bedrooms' who can spin up identical products for peanuts. The result is a market where valuations lose their anchor because differentiation via code is no longer sustainable.

From Democratization to Control

The essay contrasts the original vision of the internetโ€”a democratized platform owned by no oneโ€”with the current AI paradigm, which Priestley describes as owned and controlled by a few tech barons. He argues that AI is not a tool but a 'surveillance and data play' designed to create dependence. By convincing users that AI is an operating system for everything, incumbents are diluting critical thinking and skill. The author explicitly critiques figures like Tim Sweeney of Epic Games for pushing AI into Unreal Engine, suggesting that such moves ignore the reality that 'Jimmy in his bedroom can clone it for free.' Adobe is cited as another example of a platform already cloned by AI-driven alternatives, with copyright disputes serving as a smokescreen for the underlying loss of value.

Key Takeaways

  • Generative AI has significantly lowered the barrier to replicating software products, eroding traditional competitive moats.
  • Venture capital is trapped in a 'sunk cost fallacy,' continuing to fund AI infrastructure despite diminishing returns and rapid commoditization.
  • The author argues that AI represents a shift from the internet's democratized ethos to a model of centralized control and data surveillance.
  • Major players like Adobe and Epic Games face existential threats from low-cost, AI-generated clones that bypass proprietary tooling.

The Bottom Line

If code is cheap, the value shifts to distribution, trust, and unique dataโ€”but for most dev tools, that's a race to the bottom. Builders need to stop betting on 'better code' and start betting on things AI can't fake.